Chapter 1. Introduction
2.2 All-nanomat lithium-ion batteries: A new cell architecture platform for ultrahigh energy
2.2.3 Result and discussion
2.2.3.1 Structural uniqueness and electrochemical superiority of nanomat Si anodes 41
The nanomat Si anode is characterized by sandwich-like trilayers. A schematic representation depicting the stepwise fabrication of the nanomat Si anode is shown in Figure 2.26. First, the bottom layer was produced by concurrent electrospinning (for PAN nanofiber)/electrospraying (for SWCNT) through two different nozzles. Then, the Si active layer was introduced on the bottom layer using the same concurrent electrospinning (for PAN nanofiber)/electrospraying technique (for Si nanoparticle/SWCNT/
PEDOT:PSS). Here, Si nanoparticles were used instead of bulk Si powders to minimize the pulverization-triggered side effects.[21] Finally, the top layer, which has the same materials, composition ratio and thickness as the bottom layer, was fabricated on top of the Si active layer using the same
- 42 -
electrospinning/electrospraying process, leading to a self-standing and flexible nanomat Si anode (see the photograph in Figure 2.26).
Figure 2.26 Schematic representation depicting the stepwise preparation process (bottom layer → Si active (middle) layer → top layer) for the nanomat Si anode, along with a photograph. Each layer of the nanomat Si anode was fabricated by concurrent electrospinning and electrospraying through two different nozzles.
Figure 2.27 (a) Cross-sectional SEM image of nanomat Si anode (Si active layer (~ 10 mm) sandwiched between thin (~ 1 mm) electroconductive top/bottom layers). (b) SEM image of Si active layer, where the Si nanoparticles were compactly embedded in close contact with the PAN nanofiber/SWCNT-interweaved heteromat. (c) SEM image of SWCNT/PAN-interweaved top and bottom layers. (d) Cross-sectional SEM image of conventional Si anode with a similar Si mass loading. The conventional Si anode consisted of an electrochemically active layer (Si nanoparticles/PAA-CMC binder/carbon black = 70/20/10 (w/w/w), thickness ~ 8 μm) on top of a Cu current collector (~ 20 μm).
A cross-sectional scanning electron microscopy (SEM) image (Figure 2.27a) exhibits the formation of sandwich-like trilayers (i.e., the Si active layer (thickness ~ 10 μm) sandwiched between the thin (~ 1 μm) electroconductive top/bottom layers) in the nanomat Si anode. Figure 2.27b shows that the Si nanoparticles in the middle layer are compactly embedded in close contact with the PAN
- 43 -
nanofiber/SWCNT-interweaved heteromat skeletons, resulting in the formation of well-interconnected electron transport pathways and porous channels (that will be filled with liquid electrolytes to activate the charge/discharge reaction).
Figure 2.28 TGA profiles used to estimate the composition ratio of the nanomat Si anode. The composition ratios of the Si active (middle) layer and top/bottom layers were estimated to be (Si nanoparticles/SWCNT)/PAN = (60/6)/34 (w/w/w) and SWCNT/PAN = 40/60 (w/w), respectively. The slight increase in the weight of the Si nanoparticles, Si active (middle) layer, and nanomat Si anode above 700°C was ascribed to the formation of new silicon oxides.
From the thermogravimetric analysis (TGA) measurement (Figure 2.28), the composition ratios of the Si active (middle) layer and top/bottom layers were estimated to be (Si/SWCNT)/PAN = (60/6)/34 (w/w/w) and SWCNT/PAN = 40/60 (w/w), respectively. Meanwhile, the PEDOT:PSS, which was used as an electroconductive dispersing agent[7] for SWCNTs, was hardly detected in the SEM image and TGA profiles because of its negligibly small amount (the initial concentration of the PEDOT:PSS in the Si/SWCNT/PEDOT:PSS suspension solution was 0.03 wt%). No inclusion of the PEDOT:PSS resulted in a poor dispersion state of the (Si/SWCNT) suspension solution (Figure 2.29).
Figure 2.29 Photograph showing the poor dispersion state of (Si nanoparticle/SWCNT) suspension solution without a PEDOT:PSS additive.
- 44 -
The thin (SWCNT/PAN-interweaved) top/bottom layers in the nanomat Si anodes (Figure 2.27c) was introduced as a type of mat-based flexible electroconductive cushion. A similar concept of the trilayer electrode structure incorporating carbon nanofiber bucky papers as pseudo upper/bottom current collectors was previously reported in the development of metallic current collector-free LiNi0.5Mn1.5O4
(LNMO)-based cathodes.[22] To highlight the structural uniqueness of the nanomat Si anode, a conventional Si anode with a similar Si mass loading (~ 1.5 mg cm-2) was prepared using a typical slurry casting method. Figure 2.27d shows that a mixture layer (Si nanoparticles/PAA-CMC binder/carbon black = 70/20/10 (w/w/w), thickness ~ 8 μm) was formed on top of a copper (Cu) current collector (~ 20 μm), revealing that the conventional Si anode is thicker than the nanomat Si anode mainly due to the presence of the thick Cu current collector. Figure 2.30a shows that the nanomat Si anode exhibits remarkably higher (in-plane) electronic conductivity (specifically, top/bottom layers = 72.0 S cm-1, Si active (middle) layer = 47.5 S cm-1) than the conventional Si anode (= 0.1 S cm-1), verifying the establishment of highly reticulated SWCNT electronic networks. Meanwhile, the heteronanomat-based porous structure, in combination with the polar PAN nanofibers, allowed for better electrolyte wettability (Figure 2.30b) than the conventional Si anode.
Figure 2.30 (a) Comparison of the (in-plane) electronic conductivities of the nanomat Si anode and conventional Si anode. (b) Comparison of the electrolyte (1.3 M LiPF6 in EC/DEC = 3/7 (v/v) with 10 wt% FEC) wettability between the nanomat Si anode and conventional Si anode.
Driven by the 1D building elements-interweaved heteronanomat skeletons, the nanomat Si anode can withstand mechanical deformation. The change in the electronic resistance of the nanomat Si anode was monitored as a function of the longitudinal compression cycle (bending radius = 5 mm, deformation rate = 20 mm min-1, Figure 2.31a). The electronic resistance of the nanomat Si anode remained fairly constant even after 300 bending cycles, whereas a gradual increase in the electronic resistance was observed at the conventional Si anode. This excellent mechanical flexibility was confirmed by the SEM images after the bending cycle test (Figure 2.31b and 2.31c). No significant cracks or defects were
- 45 -
found at the nanomat Si anode, in contrast to the conventional Si anode which was seriously ruptured.
The aforementioned results demonstrate that the heteronanomat skeleton-mediated electrode architecture is effective in achieving well-developed 3D-bicontinuous ion/electron transport pathways and mechanical flexibility in the Si anodes.
Figure 2.31 (a) Change in the electronic resistance of nanomat Si and conventional Si anodes as a function of longitudinal compression cycle (bending radius = 5 mm, deformation rate = 20 mm min-1). The inset shows a photograph of the deformed nanomat Si anode. SEM images of (b) nanomat Si anode and (c) conventional Si anode after the bending cycle test (300 cycles).
Based on an understanding of the structural/physicochemical properties of the nanomat Si anodes, their electrochemical performance was investigated using a coin half cell (Si anode/PE separator/Li metal).
The nanomat Si anode exhibited larger lithiation/delithiation capacities during the 1st cycle (at charge/discharge current density = 0.05 C/0.05 C) compared to the conventional Si anode (Figure 2.32a).
The larger delithiation capacities (2,431 vs. 2,060 mAh gSi-1) of the nanomat Si anode may arise from the contribution of the SWCNTs. A blank electrode (SWCNT/PAN = 40/60 (w/w), without Si nanoparticles) underwent non-faradaic electrochemical reaction and presented an appreciable level of discharge capacity (= 232 mAh gSWCNT-1, Figure 2.32b). Similar behavior in the CNT-assisted increase in electrode capacity was already reported in our previous work.[19] Meanwhile, the nanomat Si anode
- 46 -
presented a larger delithiation capacity than the Si active (middle) layer (Figure 2.32c), indicating that the thin (~ 1 μm) top/bottom layers acted as a nanomat-based electroconductive cushion that could accommodate the lithiation/delithiation-induced volume change of the Si materials.
Figure 2.32 Initial charge (lithiation)/discharge (delithiation) profiles at charge/discharge current densities of 0.05 C/0.05 C, expressed as capacity per Si mass (= mAh gSi-1). The coin half cell (Si anode/PE separator/Li metal) was used. (a) Nanomat Si anode vs. conventional Si anode. (b) Blank electrode (SWCNT/PAN = 40 / 60 (w/w), without Si nanoparticles). (c) Si active (middle) layer.
To develop practically meaningful high-energy LIBs, considerable attention should be paid to the capacity per mass of the electrode sheets[19,23,24] rather than the capacity per mass of electrode active materials which have been widely adopted for describing the electrochemical properties of the electrode active materials themselves. Figure 2.33a shows that the nanomat Si anode presented a substantially higher (initial) delithiation capacity (= 1,166 mAh ganode-1) than the conventional Si anode (= 122 mAh ganode-1) when our view switched from the capacity (mAh gSi-1) per mass of Si material to the capacity (mAh ganode-1) per mass of Si anode sheet. This improvement in the electrode capacity results from the architectural uniqueness of the nanomat Si anode. In particular, the removal of a heavy Cu foil current collector is crucial for lowering the areal weight of the nanomat Si anode (Figure 2.33b), eventually
- 47 -
exerting a beneficial effect on the electrode sheet-based capacity.
In addition, the nanomat Si anode exhibited a higher delithiation rate capability than the conventional Si anode (Figure 2.34a). This result was further verified by galvanostatic intermittent titration technique (GITT) analysis.[25,26] Figure 34b shows that the nanomat Si anode effectively suppressed the rise in cell polarization upon repeated current stimuli (at a current density of 0.2 C and interruption time between each pulse of 60 min), wherein the internal cell resistances were presented as a function of the state of charge (SOC) and depth of discharge (DOD). The superior rate performance of the nanomat Si anode was ascribed to the heteronanomat-enabled, 3D-bicontinuous ion/electron transport channels.
Figure 2.33 (a) Initial charge (lithiation)/discharge (delithiation) profiles of Si anodes at charge/discharge current densities of 0.05 C/0.05 C, expressed as capacity per mass of Si anode sheet (= mAh ganode-1). (b) Comparison of the areal masses (= mg cmAnode-2) between the nanomat Si anode and conventional Si anode.
Figure 2.34 (a) Delithiation rate capability (expressed as mAh ganode-1) of half cells (= Si anode/PE separator/Li metal). The discharge current density was varied from 0.2 to 3.0 C at a constant charge current density of 0.2 C.
(b) GITT profiles of half cells upon repeated current stimuli (at a current density of 0.2 C and an interruption time between each pulse of 60 min) and variation in the internal cell resistance as a function of SOC and DOD.
A major challenge facing Si anodes is how to maintain their structural integrity and ion/electron conduction pathways during repeated lithiation/delithiation reactions, which often cause changes in the
- 48 -
volume of the Si anodes. The overpotential distribution of the Si anodes in the through-thickness direction was quantitatively examined as a function of SOC (Figure 2.35a). For this measurement, Cu taps were attached to both the top and bottom sides of the Si anodes (used as a working electrode), and a lithium metal foil was used as a counter/reference electrode (Figure 2.35b). During the lithiation reaction (at a current density of 0.2 C), the nanomat Si anode exhibited a smaller voltage gap (ΔV (=
Vtop – Vbottom, at 50% SOC) = 1.0 mV) than the conventional Si anode (ΔV = 6.5 mV), indicating lower cell polarization in the through-thickness direction. This result demonstrates that the well-developed ion/electron conduction pathways of the nanomat Si anode facilitated the lithiation kinetics, leading to a uniform voltage distribution in the through-thickness direction.
Figure 2.35 (a) Overpotential distribution of Si anodes in the through-thickness direction as a function of SOC.
The voltage gap (DV (= Vtop – Vbottom, at 50% SOC) was measured at a discharge current density of 0.2 C. (b) Schematic representation of a pouch cell designed for the in situ EIS measurement of the overpotential distribution of Si anodes in the through-thickness direction (i.e., voltage difference between the top and bottom sides).
The cycling performance (expressed as mAh gAnode-1) of the nanomat Si anode was examined at charge/discharge current densities of 0.2 C/0.2 C (Figure 2.36). Remarkable improvement in the capacity retention with cycling (= 87% after 100 cycles) was observed in the nanomat Si anode compared to the conventional Si anode that exhibited a sharp capacity decay (~ 0% after 100 cycles).
These cycling results were further verified by analyzing the (through-thickness directional) structural
- 49 -
change of the Si anodes. Figure 2.37a and 2.37c show that the thickness increase in the Si anodes after 100 cycles was significantly suppressed at the nanomat Si anode (12 (before cycling) 21 (after 100 cycles) μm) compared to the conventional Si anode (8 45 μm). Moreover, for the nanomat Si anode, neither appreciable cracks nor morphological disruptions were observed after 100 cycles, in contrast to the conventional Si anode (Figure 2.37b and 2.37d). This result demonstrates that the heteronanomat sheet architecture effectively mitigated the volume change of the Si anodes and maintained the 3D- bicontinuous ion/electron channels during repeated cycling.
Figure 2.36 Cycling performance (expressed as mAh gAnode-1) of half cells: nanomat Si anode vs. conventional Si anode. The half cells were cycled at charge/discharge current densities of 0.2 C/0.2 C under a voltage range of 0.01-1.2 V.
Figure 2.37 SEM image after the cycle test (100 cycles). (a) and (b) Nanomat Si anode. (c) and (d) Conventional Si anode. The thickness change (Δt = thickness (after 100 cycles) – thickness (before cycling)) of the Si anodes was recorded.
- 50 -
2.2.3.2 Nanomat OLO cathodes based on rambutan-shaped OLO/MWCNT nanocomposite powders
Figure 2.38 SEM image of rambutan-shaped OLO/MWCNT nanocomposite (referred to as “R-OM”) powders.
The inset is a photograph of rambutan fruit.
A formidable problem in OLO cathode materials is their low electronic conductivity. Here, we prepared a new class of nanocomposite powders composed of OLO nanoparticles and multi-walled carbon nanotubes (MWCNTs) to address this issue. The bare OLO powders (average diameter = 5 μm, Figure 2.39a) were mixed with the MWCNTs by ball milling in the presence of 1 wt% PVP, yielding rambutan- shaped OLO/MWCNT (= 100/10 (w/w)) nanocomposite (referred to as “R-OM”) powders (Figure 2.38).
Notably, the highly reticulated MWCNT networks in the R-OM powders are expected to provide a significant improvement in electronic conductivity compared to the bare OLO powders, which will be further discussed below.
Figure 2.39 (a) SEM image of bare OLO powders (average diameter = 5 mm). (b) SEM image showing the poor mixing state of OLO nanoparticle aggregates and MWCNTs. (c) Viscosity of the OLO/MWCNT suspension (with PVP vs. without PVP). (d) Viscoelastic properties (G’ and G”).
- 51 -
PVP, which is known as a nonionic dispersant for MWCNTs, was used to stabilize the dispersion state of the OLO/MWCNT suspension. As a control sample, OLO/MWCNT (= 100/10 (w/w)) nanocomposite powders were prepared in the absence of PVP. In the control powders, the OLO nanoparticle aggregates were not homogeneously mingled with the MWCNTs (Figure 2.39b). This PVP-assisted stabilization of the OLO/MWCNT suspension was further verified by analyzing its rheological properties (Figure 2.39c and 2.39d).[27,28] The OLO/MWCNT suspension containing the PVP dispersant exhibited a higher shear viscosity and more elastic behavior (i.e., G’ (storage modulus)
> G” (loss modulus)) over a wide range of shear stress compared to the OLO/MWCNT suspension without PVP.
Figure 2.40 (a) N2 adsorption-desorption of R-OM powders. The inset shows the HR-TEM image. (b) BJH plot showing the average pore size of the R-OM powders.
The good dispersion state of the OLO/MWCNT suspension allowed for the formation of highly interconnected MWCNT networks/porous interstitial voids (that serve as ionic conduction channels) in close contact with OLO nanoparticles. The porous interstitial voids of the R-OM particles were verified by a high-resolution transmission electron microscopy (HR-TEM) image (inset of Figure 2.40a). The N2 adsorption-desorption isotherm (Figure 2.40a) shows that the specific surface area of the R-OM powders is 22.2 m2 g-1. In addition, the BJH plot (Figure 2.40b) showed that the average pore sizes of the R-OM powders were estimated to be 24.0 from the desorption mode and 33.1 nm from the adsorption mode. The porous interstitial voids, in collaboration with the MWCNT networks, are expected to facilitate the redox reaction kinetics of the R-OM powders.
The effect of MWCNT content on the morphology and properties of the R-OM powders was investigated. Randomly aggregated OLO nanoparticles were observed in the absence of MWCNTs (Figure 2.41a). Over the entire range of MWCNT contents (= 3, 5 and 15 wt%, shown in Figure 2.41) examined herein, rambutan-shaped, micron-sized OLO/MWCNT nanocomposite powders were formed, indicating that the MWCNTs act as an electroconductive fibrous binder that tightly holds the OLO
- 52 -
nanoparticles. For MWCNT content of 3 and 5 wt%, the population density of the MWCNTs was not sufficiently high for constructing well-reticulated MWCNT networks. Meanwhile, an excessively high MWCNT content (= 15 wt%) resulted in the generation of independently existing OLO and MWCNT aggregates besides the OLO/MWCNT nanocomposite powders. To quantitatively elucidate the effect of the MWCNT content in the R-OM powders, OLO cathode sheets (R-OM powder/PVDF binder/carbon black additive = 80/10/10 (w/w/w)) were prepared using a typical slurry casting method.
The (in-plane) electronic conductivities of the OLO cathodes increased with MWCNT content (Figure 2.42a). The discharge rate capability of the OLO cathodes (at a constant charge current density of 0.2 C) was examined using a coin half cell (OLO cathode/PE separator/lithium metal anode). The R-OM powders with the 10 wt% MWCNT showed the highest rate performance (Figure 2.42b). The lower rate capability of the 15 wt% MWCNT compared to the 10 wt% MWCNT could be attributed to the unwanted formation of the independently existing OLO and MWCNT aggregates, which are unfavorable for ion/electron transport in the R-OM powders. Based on these results, the optimal MWCNT content of the R-OM powders was suggested to be 10 wt%.
Figure 2.41 SEM images of R-OM powders with different MWCNT contents: (a) 0, (b) 3, (c) 5, (d) 15 wt%.
The nanomat OLO cathode includes the PAN/PVP mixture nanofibers as 1D-shaped binders. PVP, due to its pyrrolidone ring (containing lone pair electrons)-driven Lewis basicity,[29,30] is known to scavenge metal ions (e.g., Mn2+) dissolved from lithium metal oxide-based cathode materials. Here, PAN and PAN/PVP (= 50/50 (w/w)) films were prepared as model systems to quantitatively examine beneficial function of PVP. No bare PVP film was tested because PVP tends to lose its dimensional stability easily upon being immersed in LIB electrolytes.[20] The Mn2+-chelating ability of the PAN and PAN/PVP films in the electrolyte solution was analyzed by measuring the amount of captured Mn2+ ions (Figure 2.43),
- 53 -
where the PAN and PAN/PVP films were immersed in manganese perchlorate solution (10 mM Mn(ClO4)2-containing 1.0 M LiPF6 in EC/DMC = 1/1 (v/v)) for 2 h at room temperature. The amount of Mn2+ ions (= 224 ppm) trapped by the PAN/PVP film was significantly larger than for the PAN film (= 3 ppm). The beneficial effect of this Mn2+-chelating capability of the PAN/PVP on cell performance will be discussed in the following section.
Figure 2.42 (a) (In-plane) Electronic conductivities and (b) discharge rate capabilities (coin half cell (OLO cathode/PE separator/lithium metal anode, the discharge current density was varied from 0.2 to 5.0 C at a fixed charge current density of 0.2 C) of the OLO cathodes (R-OM powder/PVdF binder/carbon black additive = 80/10/10 (w/w/w), where the MWCNT contents in the R-OM powders were 0, 3, 5, 10 and 15 wt%.
Figure 2.43 Amount of Mn2+ ions captured by PAN and PAN/PVP (= 50/50 (w/w)) films (measured by ICP-MS analysis).
- 54 -
Figure 2.44 Conceptual representation depicting the structural uniqueness of the nanomat OLO cathode. Its photograph was inserted.
After preparing the major components, the nanomat OLO cathodes were fabricated through concurrent electrospinning (for PAN/PVP nanofibers) and electrospraying (for R-OM powders/MWCNTs). A conceptual representation depicting the structural uniqueness of the nanomat OLO cathode, alongside a photograph showing the mechanical flexibility, is shown in Figure 2.44.
Figure 2.45 TGA profiles used to estimate composition ratio of the nanomat OLO cathode; the weight-based ratio of OLO/MWCNT/(PAN-PVP) was 65/13/22 (w/w/w).
The composition ratio of the nanomat OLO cathode was estimated to be OLO/MWCNT/(PAN-PVP) = 65/13/22 (w/w/w) from the TGA measurement (Figure 2.45). Figure 2.46a shows that the R-OM powders were densely packed in the MWCNT/(PAN/PVP)-interweaved heteromat without a metallic Al foil current collector, leading to the formation of 3D-bicontinuous ion/electron transport pathways, which appeared similar to the results for the nanomat Si anode. The highly interconnected electronic channels of the nanomat OLO cathode were confirmed by comparing its electronic conductivity with the conventional OLO cathode. The conventional OLO cathode (OLO/PVdF/carbon black = 80/10/10 on an Al foil, Figure 2.46b) was fabricated using a slurry casting method. Figure 2.46c shows that the